In the next three years, the number of next-generation sequencing (NGS) samples processed will more than double. The NGS protocol will need to be industrialized, which will require identifying and eliminating bias and errors in assays. Major productivity gains, including reducing systematic bias, can occur by improving sample handling during sample preparation. All of the major technology platforms rely on ligation of DNA fragments during sample prep. This step is a major contributor to system-wide bias.
Three primary types of errors lead to bias: systematic, genome coverage, and batch processing effects. Depending on your sequencing platform, the genome content in your sample and data may be susceptible. Systemic bias is the most difficult to identify and therefore the most difficult to correct. It is pernicious because it can remain regardless of the genome coverage and platform.
Low-quality chemistry and poor liquid handling may exacerbate errors but can be minimized.
Chemistry is the main culprit behind erroneous base calls. Sequencing reactions are quite precise; minute variations in reagent volumes, flow, and temperature can lead to base substitution errors or an incomplete sequence extension. Because all technology platforms use adapters, base substitution or incomplete sequence extension can also mean fragment loss in the library. Eliminating these errors is vital when using an amplification-based protocol, as they are magnified during each cycle.
One way to minimize bias is to improve sample preparation and handling. Target enrichment, library preparation, and library amplification are key steps that are especially susceptible to introducing bias. At each step of enzymatically modifying DNA into genomic DNA (gDNA), the fragment diversity can be compromised. At the sample prep and quality control (QC) stages, input DNA can be twice the mass needed to run in the sequencing machine; in the final library, only a small fraction of the original sample is still present.
Platform evaluation studies have demonstrated systematic bias unique to each manufacturer’s system. Many recent studies have given attention to correcting bias resulting from amplification in GC-rich regions. These studies did not find ways to predict bias, but they did characterize it by platform. A study by the Sanger Institute noted that sequences generated by the Ion Torrent, MiSeq, and PacBIO RS II platforms all perform “at a near-perfect level” for accurately identifying GC-rich regions.
Calling the opposite strand, AT rich regions, was more challenging. When using platforms dependent upon PCR amplification, researchers had challenges calling neutral and moderately rich AT regions.
Systematic bias can also occur in the post-processing stage during mapping and alignment. Comparing benchtop sequencers, a study by the Centre for Systems Biology in Birmingham, U.K., documented bias in calling insertions and deletions (indels).
The root cause is yet to be found whether the bias resulted from sequencing and chemistry or if it was generated during data filtering and variant calling.
Limiting Exposure to Bias by Adopting Automation
One effective way to minimize error and avoid systematic bias is to adapt user-friendly liquid-handling protocols. Automating liquid-handling steps, especially during sample preparation, simplifies the overall processes. It reduces sample-to-sample variability as well as technician-to-technician variability, minimizing batch-processing bias. Most importantly, it ensures that you are operating in a precision chemistry environment and minimizing systematic bias.
Precision liquid handling improves the accuracy of passive and active chemistry steps within the sample-prep workflow. For example, precision air dispensing for steps such as the end-repair reaction or the critical adaptor ligation steps enable technicians to dispense the exact amount of reagents needed—no more, no less.
Hamilton Robotics has a unique technology for aspirating and dispensing liquid. Hamilton’s Compressed O-Ring Expansion (CO-RE™) technology and monitored air-displacement (MAD) pipetting software enable pipette tips to attach and detach with a minimum amount for pressure and agitation (Figure 1). During the passive chemistry, or washing steps, agitation is coming from the peltier—not from pipette tips changing. All Hamilton workstations use the CO-RE technology (Figure 2).
Researchers need not think that automation has to be outside their budget. If a lab is processing more than two 96-well plates per week, the savings from using fewer reagents will typically pay for the instrument.
How Else Can Your Lab Benefit from Automation?
Automating the preanalytic phase can bring productivity gains to next-generation sequencing service laboratories. Scaling up to industrial levels will require the consistency and tracking that only automation can provide.
Automation enables sample-prep runs to happen overnight, so gDNA can be ready for the sequencer in the morning. Existing protocols only need minimal modifications, and engineers from automation manufacturers often help your lab migrate from manual to automated processes.
In addition, sample-tracking software provided in most liquid-handling systems are 21 CFR 11 compliant, which requires all clinical labs to demonstrate secure sample handling. The tracking software documents chain of custody as patient samples are transported to the laboratory and through the assay workflow. These audit trails satisfy documentation requirements for quality checks and ensures that patient confidentiality controls are in place.
Conclusion and Future Trends
NGS technologies hold much promise for advancing diagnostics and making therapeutic decisions, but clear guidelines are needed for quality control procedures, reporting results to patients, certified reagents, assay validation, and proficiency testing.
Labs using the same instrumentation may be using different sample-prep kits, different reagents and different algorithms to call variants. Given the broad and varied community using these technologies, it will likely be difficult to gain consensus on how to perform a test from end to end. This summer the American College of Medical Genetics published guidelines for clinical NGS and attempted to establish standards.
Automation for sample prep is a way to solve many of these challenges while improving productivity and saving reagent costs. As NGS assays become more routine, variability from chemistry and manual steps need to be documented, if not eliminated. To ensure your system is taking advantage of these advancements it is important to understand how variation and systematic bias is affecting your results.